Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Transportation

Model S Touchscreen Failures Prompts NHTSA Investigation (theverge.com) 103

mschaffer writes: Some older Teslas (Model S) are experiencing problems due to a worn-out flash eMMC chip. The loss of touchscreen functionality prevents operators from being able to use the backup camera. This may potentially impact 63,000 cars. The agency also acknowledges that the same chip was used in the 2012 to 2018 versions of the Model S and the 2016 and 2018 version of the Model X, accounting for around 159,000 vehicles.

"The agency currently has two other such investigations open into Tesla," adds The Verge. "One, opened last October, is about an over-the-air software update Tesla issued that was meant to limit a possible fire risk in the battery packs of the company's cars. The other, opened in January, is looking into claims of Tesla's vehicles undergoing 'sudden unintended acceleration.'"
This discussion has been archived. No new comments can be posted.

Model S Touchscreen Failures Prompts NHTSA Investigation

Comments Filter:
  • by Joe_Dragon ( 2206452 ) on Wednesday June 24, 2020 @06:38PM (#60224380)

    mouser price $11-$50 Our price
    $100 +
    Install $100
    Software reload $100
    Shop fees $30
    and we need 4 hours.

    • by dgatwood ( 11270 )

      Nope. The eMMC is soldered onto the board.

      • Then we need law saying that disks must be on cards / drives that can be swapped and end users must have access to the software image with no dealer only pairing needed under the law

        • by dgatwood ( 11270 ) on Wednesday June 24, 2020 @07:02PM (#60224478) Homepage Journal

          Yes, we do. I have the same objection to the entire Mac platform for the past couple of years. Storage is a consumable, and for a consumable to not be replaceable means that the product is disposable. That's bad. In a car, it's unconscionable.

          On the flip side, in a car, there's a lot of vibration, so there's a potential reliability tradeoff involved in making it easier to replace that part. I'm honestly not sure what the right solution is in the automotive world.

          The real problem is that Tesla seriously cheaped out on the eMMC. They should have used a part that was massively larger than what they actually needed, so that they could do simple static wear leveling of unused blocks and avoid making the part fail prematurely. Instead, the two copies of the firmware take up the bulk of the chip's capacity, leaving precious little space for writing logs, so they would have to constantly do dynamic wear leveling involving periodically moving the blocks that contain the operating system itself to new pages (risky) to avoid excessive wear on the remaining flash pages.

          This could all have been avoided if they had stuck 64 GB of flash into the thing, but 8 GB just doesn't cut it. It was woefully too small long before they finally updated it to 32 GB with the release of MCU 2 hardware. They should have revved the chip to a much larger part many years earlier than they did. In fact, the third-party folks rebuilding Tesla MCUs are replacing the 8GB chips with larger parts precisely for that reason.

          The unfortunate reality is that this is going to eventually be a problem for MCU2 cars as well, and there are a lot more of those. Even the 32 GB parts in MCU2, IMO, aren't nearly big enough, given how much these computers have to store on those parts. But at least those are likely to last more than single-digit years... assuming they don't add too much to the firmware image.

          • by drinkypoo ( 153816 ) <drink@hyperlogos.org> on Wednesday June 24, 2020 @07:12PM (#60224514) Homepage Journal

            I'm honestly not sure what the right solution is in the automotive world.

            Usually a bigger connector, and some sort of finger pressing down on the component so it doesn't shake itself loose.

            The real problem is that Tesla seriously cheaped out on the eMMC. They should have used a part that was massively larger than what they actually needed, so that they could do simple static wear leveling of unused blocks and avoid making the part fail prematurely.

            Yep. That would have cost basically nothing compared to what it's going to cost to repair all of these boards. And it's not unlikely that is what will actually happen so that they don't need to replace them all, which would be even more expensive.

            For the record though, the hardware in a car doesn't have to be massively more durable than other mobile hardware like a cellphone, which has to be able to stand up to some pretty huge shocks. Even laptops are expected to take some very high Gs these days. And the car computers are frequently mounted on soft bushings.

            • by dgatwood ( 11270 )

              Usually a bigger connector, and some sort of finger pressing down on the component so it doesn't shake itself loose.

              Ostensibly, yes, but you're talking about a high-speed data interconnect. Those typically require balanced signal wires close together. I'm not sure how well that would work.

              Either way, what I meant was that I'm not sure how you could make it so that replacing it wouldn't involve a custom board that you can only get from the manufacturer. If you still have to have custom hardware to extract

              • you can get m.2 cards and they don't need to be soldered down to be fast and does an car really need an disk that fast? or why can't they log to an slower card that can be at SD card speeds.

                • They were probably doing something insane like swapping, or had some bug that caused excessive writes.

                  Either way, some kind of M.2 storage is probably the best answer here, with a top-quality connector and a retention screw with thread locking compound on it. Whether it's SATA or PCIE -- and if the latter, one lane or four -- the benefits of using field-replaceable storage are massive. Apple can afford to jerk their users around, Tesla can't.

                  • The odds of their being a SATA or pci-e bus is pretty low. eMMC is just a serial signalling protocol, much like SD.

                    SD or its newer successors are plenty fast enough, and the connectors can be reliable enough to do this no problem. No need to think this is a PC."

                    They just got sloppy and "did not do the math", or they hit a wear amp issue on the eMMC device that did not show up in testing. eMMC parts have their equivalent of SMART, so they had the data. I bet they decided to store/buffer a video stream an

                    • by dgatwood ( 11270 )

                      According to Wikipedia, eMMC is an 8-bit parallel interface. That means, among other things, that trace length likely matters.

                    • by tap ( 18562 )
                      This is very much true. eMMC can operate in 4-bit and 8-bit mode, but of course one tries for 8-bit.

                      At the higher speeds the signal integrity requirements are quite a challenge, even for a chip soldered on the board. The trace lengths must be matched closely and routing across the board or over a via through layers is problematic. I've never seen a socketed eMMC chip.

                      The original eMMC bus speed was quite a bit slower, 26 MHz, and more suited to the parallel interface with separate clock. IMHO, tt's

                  • by Octorian ( 14086 )

                    They were probably doing something insane like swapping, or had some bug that caused excessive writes.

                    As I understand it, they used to do excessive logging on the vehicle. This caused excessive writes, which obviously can wear out the eMMC.

                    Also, the eMMC is supposedly on a removable module that's attached to the MCU (the big touchscreen component). There are whole forum threads about how to remove it and replace the chip, and even people offering help with that as a service. However, this is usually done when its going flaky but hasn't completely failed yet.

            • The hardware is often the same, with slightly larger temperature tolerance, then they just pot it in epoxy and call it a day.
              • Actually it needs to be far more EMI and power spike tollerant. "Automotive grade" frequently means little more than "can be run wait out of spec for a very brief period"

                • Psshhtt. Throw conductive powder in your potting compound and leave a few ground planes on there with no mask, it’s good emnuf. All but the dirt cheapest boards have an input low pass filter and transient suppression on exposed data lines these days. Automotive typically function from -40 to 125C while commercial is more like 0-85 but yea, I wouldn’t hold my breath that it’s gonna keep working for long at either extreme.
            • by tap ( 18562 )
              It's really the temperature range that matters for automotive qualified ICs. It's 70C for consumer and 125C for automotive.
            • by AmiMoJo ( 196126 )

              For automotive use they make special automotive grade SD cards with a wider than normal temperature range and a full factory test. They also do solder down versions.

              Oh yeah, if you didn't know consumer grade SD cards don't get a full flash memory test. They just over-provision a little and rely on the defect/return rate being low. For automotive the cost of replacement is high so they actually do more than a basic functional test.

              The real fix though is to reduce the amount of logs written to the memory. Tha

          • As it turns out, automotive engineers are pretty good at making sure shit doesn't shake itself apart. We've had electronics in cars for decades that don't shake parts off.

          • by raymorris ( 2726007 ) on Thursday June 25, 2020 @12:46AM (#60225262) Journal

            >> end users must have access to the software image with no dealer only pairing needed under the law

            > Yes, we do

            I don't buy laptops that don't have replaceable storage and ram.

            Also, I don't want random people hacking around in the Tesla autopilot to make the car do "cool stuff". Wanna root your phone? That's cool; I rooted mine. Not so cool to download a hacked Tesla firmware from HackForums.com and then put it in control of a 4,300 pound car that's barreling toward me at 75 mph.

            To me it's not entirely clear where lines should be drawn sometimes.

          • by AmiMoJo ( 196126 )

            They should have used reduced the amount of logging going on in production vehicles too. A lot of it is to gather data for their self-driving car programme and to help put the blame on customers when things go wrong. Naturally only Tesla has access to these logs, the """owner""" of the car doesn't.

            Since it's a design defect they basically have to put it right for free in the UK but elsewhere you may be SOL.

        • Tesla doesn't have dealers.
      • by laird ( 2705 ) <lairdp@gm a i l.com> on Wednesday June 24, 2020 @07:10PM (#60224512) Journal

        There's no problem with the eMMC being soldered to the board - that's pretty routine for embedded electronics, as it's more reliable. The problem is that Tesla's software started logging more and more to the eMMC so that it caused the chip to hits its lifetime capacity far sooner than originally planned. This isn't a new issue - it was raised, and supposedly addressed by changes to Tesla's software to do "wear leveling" and to log less aggressively, a year or two ago: https://insideevs.com/news/376... [insideevs.com] . Tesla can replace the media controller, and there are several third-party repair options (de-soldering and replacing the eMMC) that cost much less than Tesla's board swap.

        • by dgatwood ( 11270 )

          Logging certainly exacerbates the problem, but the eMMC also appears to be used for map tile caching, and that puts a decent amount of wear on it, too, particularly given that most of the space contains static data (the firmware binary). That's basically worst-case abuse as far as wear leveling goes.

          • particularly given that most of the space contains static data (the firmware binary). That's basically worst-case abuse as far as wear leveling goes.

            Depends on the wear levelling system inside the eMMC's embed controller.

            Some constructors even do "static" wear levelling. I.e.: the static data is moved around occasionally, which as the double benefit of:
            - avoiding decay (remember a few years back Samsung 840 EVO's voltage drop causing slow downs on static data and which necessitated a firmware upgrade enabling exactly this ?)
            - increase the spread of write even if an eMMC contains a lot of static data (exactly the situation you describe).

            (Transcend even a

            • by dgatwood ( 11270 )

              This is whatever 8 GB part was cheap in 2012. I think we can safely assume that it does wear leveling only for modified pages. :-)

            • by tap ( 18562 )

              Depends on the wear levelling system inside the eMMC's embed controller.

              For every eMMC I've used, the manufacture considered the wear leveling proprietary and wouldn't provide any information. Even erase block and page sizes have to be guessed using benchmarks and testing.

              Is there information out there on this for eMMC? I'd love to know this.

        • by AmiMoJo ( 196126 ) on Thursday June 25, 2020 @06:09AM (#60225760) Homepage Journal

          Repair of design defects should be free. They are under UK consumer law. This is entirely Tesla's fault, no one should pay a penny except them.

      • Re: (Score:2, Insightful)

        by Barsteward ( 969998 )
        Is this yet another case where the Tesla Short Sellers are manufacturing complaints on behalf of the owners like they did with "unintended acceleration? They are still getting their pockets emptied at the moment so there is nothing like creating bad publicity for a problem where none exists. This is just normal wear and tear, dashboard electronics fail now and again on ordinary ICE vehicles too. The problems don't seem dangerous e.g. loss of rear camera
      • by TheGratefulNet ( 143330 ) on Thursday June 25, 2020 @08:09AM (#60226046)

        sometimes, it make sense to solder it on.

        connectors that have to meet auto grade standards (I work in the industry) are expensive. heat and cooling and vibration qual for auto grade is hard and it should be, you don't want things to fail. so, every metal to metal contact, you have to argue for (and I'm against them; at home, fine; but in the car, no!).

        the other reality is cost. every 5 cents they save on a cost reduction in parts matters. a group just down the hall from me spends a lot of its time cost reducing existing designs, even just removing some leds that were there for debug at bring-up or changing a part for a diff vendor one (new footprint, new layout, new qual; but 'parts savings'!).

        I don't agree with all of this, but its how the car biz works and its not going to change.

        now, that said, tesla really fucked up with this design. not the hardware per se, but the fact that they used linux and had it log (syslog, etc) to flash and the apps that run on the os also log.

        I see it in my own company, sad to say. I brought the issue up and we also mostly ignored it ;(

        but tesla has been told many times that they need to hire truly senior people to do designs and not 'freshers' from college. that's where I put the blame; cheapness on hiring and having too many 'desktop guys' try their hand at embedded.

        and guess what, embedded is NOT just 'desktop with no keyboard and screen' guys.

        tesla should pay the full cost of board replacement. they should not put sockets in - wrong answer - but they need to lower logging (they did, I'm told) but also realize HOW they are running and that they don't have lots of ram and lots of 'disk'. every write to flash has to be though out. if flash is not the right thing, use another form of persistent storage. even industrial sd-card is auto rated and that could be their cache for logfiles and such.

        putting the cost on the user when the user had nothing to do with itis a real amateur job, ELON. dammit, ELON, grow up and start running your company like an adult and not some spoiled child. or, better yet, give it up and let someone else run tesla. (ob disc, I do own a tesla car and mostly like it, but truly hate the company for how they take such a good thing and always mess something up about it.)

        tesla has limited time; there are companies close on their heels (for some values of 'close') and they won't be the only game in town, in less than 5 years time. people buy teslas since there is no competition, but that's on the move and tesla needs to make a fast change if they want to stay relevant, going forward. they have the lead NOW but in high tech, that changes quickly.

    • by AmiMoJo ( 196126 ) on Thursday June 25, 2020 @06:04AM (#60225752) Homepage Journal

      And the rest. Tesla charges $2000-3000 for an MCU replacement out of warranty. They don't replace the memory, they replace the entire computer. It's a fair bit of work.

      The other big issue that affects all cars over a certain age is yellowing. The screen goes yellow where it is glued together. Tesla will give you one free UV treatment to temporarily fix it but don't consider it a warranty issue.

      • yep, I got hit with the MCU eMMC Issue on my 2013 P85+. Last august I had to get the screen replaced for 1100.00 Now this repair will cost 1600.00 and it comes with another screen. They will not just fix the eMMC, p/n that is ordered for the repair includes a new screen that I don't need. I declined to pay for the repair at this time to wait until MCU2 is available, then a few days later my car will no longer even charge. Well it charges extremely slow to a maximum of 46 miles. They say the battery and driv
    • This is a safety/recall issue.

      Tesla will send a tech to you to replace the component free of charge.

      This is how they handle most of their problems. Since they do not run traditional dealerships, they have no incentive to make you come to them for services that can be performed in the field.

  • Comment removed (Score:5, Insightful)

    by account_deleted ( 4530225 ) on Wednesday June 24, 2020 @06:41PM (#60224392)
    Comment removed based on user account deletion
    • by Luthair ( 847766 )
      It probably doesn't help that at least in some instances Tesla hasn't used automotive grade components. (No idea if that is the case here)
      • Re: (Score:1, Interesting)

        by danskal ( 878841 )

        As far as I know some of their electronics are military grade. Much better than "automotive grade".

        Can you back your vague, FUDdy comment up with any kind of facts?

        If you watch some of Sandy Munro's videos, you'll end up thinking twice about buying anything else.

        • Re:Complexity (Score:4, Informative)

          by Hodr ( 219920 ) on Thursday June 25, 2020 @09:24AM (#60226314) Homepage

          "Military Grade" Isn't a thing. Every branch of the military has differing specifications for every type of component. Got an RF cable going into an Airplane for the AF? That's going to have different shock/vibe/heat requirements than one for the Navy (who launches from carriers and also needs components resist salt/moisture), which is also going to be different than what the FAA requires.

          And then there is the difference between products developed by the military/government and those commercially developed and allowed to be used. The first must pass full qualification testing and be built with appropriately sourced components that also meet exacting standards. The later may possibly need some qualification testing depending on the criticality of the application (going in an airplane, it's getting tested. Going in an office, nope).

          Take a standard off the shelf hard drive, has it ever been used in a device approved for the military (not a toughbook, or other ruggedized product, but lets say a PC under a desk in some office). Welp, looks like it's MILITARY GRADE!

           

          • by danskal ( 878841 )

            Thankyou, captain obvious. I was just paraphrasing Sandy Munro, who said that the technology was much more advanced than anything he had seen in the auto industry. He had only seen it in military applications. He didn't go into detail about whether he was talking about nuclear submarines, ICBMs, tanks or something else. But I'm pretty sure he wasn't talking about military office equipment.

      • Re: Complexity (Score:4, Insightful)

        by c6gunner ( 950153 ) on Thursday June 25, 2020 @02:24AM (#60225398) Homepage

        It probably doesn't help that at least in some instances Tesla hasn't used automotive grade components.

        That isn't inherently a problem. In this specific case they would probably have been far better off using a typical smartphone EMMC module for the software, plus a simple MicroSD card for the logging/map files/etc. Even the highest quality EMMC will wear out fairly quickly if you abuse it the way they did; makes more sense to just treat it better and write the non-critical stuff to some cheap, very easily replaceable storage.

        • by Luthair ( 847766 )

          That isn't inherently a problem.

          Of course its a problem to use components which aren't rated for the location they're used. Components in cars are subjected to significant temperature ranges from operating as a greenhouse in the summer to the freezing cold. If they aren't designed for those operating ranges, they aren't going to work for a decade+ (or in the case of Tesla's screens, a few years).

          • Components in cars are subjected to significant temperature ranges from operating as a greenhouse in the summer to the freezing cold. If they aren't designed for those operating ranges, they aren't going to work for a decade+ (or in the case of Tesla's screens, a few years).

            This is a common refrain, and an overblown one. You think a $150 aftermarket CD deck uses sooper-speshul automotive-grade components? I never had one of those crap out on me.

            I also kept a standard tablet (the original Acer Iconia Tab from 2011) in my car for about 6 years as a navigation and entertainment device. It stayed in the car facing everything from the scorching heat of summer in the American south to the frozen wastelands of Canadian winter. Seemed to do just fine.

            With Tesla's "always on" vehic

            • by tlhIngan ( 30335 )

              This is a common refrain, and an overblown one. You think a $150 aftermarket CD deck uses sooper-speshul automotive-grade components? I never had one of those crap out on me.

              I also kept a standard tablet (the original Acer Iconia Tab from 2011) in my car for about 6 years as a navigation and entertainment device. It stayed in the car facing everything from the scorching heat of summer in the American south to the frozen wastelands of Canadian winter. Seemed to do just fine.

              With Tesla's "always on" vehicles

              • That was an impressively comprehensive response, thanks :) however I wasn't arguing that nothing in a vehicle had to be capable of handling extreme temperature ranges; if you go back and check, the original guy said that Tesla hasn't always used "automotive grade components" and I pointed out that this isn't inherently a bad thing.

                Pretty sure we are in complete agreement here.

                • by Luthair ( 847766 )
                  So you agree that components in cars are subjected to extreme temperature ranges for years, but you repeatedly claim that it isn't a bad thing that Tesla has a history of choosing components which are not designed or tested for those temperature ranges.... Did they mail you the pom-poms or were you happy to pay for it out of pocket.
                  • So you agree that Tesla is awesome and builds fantastic cars, but you're worried that you won't be able to afford one. I get ya man, it's ok. Prices will come down, don't worry.

            • Pepperidge Farm remembah's a time when analog clocks in cars didn't last more than a year! :)

    • I wouldn't call it complexity - I would call it scalability. Each system is built of a number of sub-systems. Yes, the total complexity is mind boggling, but it all works in a hierarchical or networked fashion. The internet is unfathomably massive, yet it works well. Any given component of the internet does not need the knowledge about every detail of every other component. A modem designer doesn't need to know the design of the lasers or transistors used in the ASICs. A switch designer doesn't need to know
    • Re:Complexity (Score:5, Insightful)

      by LynnwoodRooster ( 966895 ) on Wednesday June 24, 2020 @07:06PM (#60224498) Journal

      CE drivetrains are nearly bullet-proof these days, as long as you change the fluids on schedule (typically 8K miles for oil, 50K for coolant, 100K miles for transmissions). You can easily expect to see 200K+ miles from an engine. What tends to break down before that is the suspension. Or the interior. Or a FREAKING 17" SCREEN IN THE MIDDLE OF THE DASH...

      Buttons, knobs, and sliders FTMFW!

      • Not remotely true. The DPF is a major source of headaches and costs a ton of money to replace, even on reliable vehicles like the Toyota Hilux and soon gas engines will all have a GPS (gas particulate filter). Then there are oil consumption problems related to the industry-wide effort to reduce consumption and CO2 emissions even by a little. And that's before all the plasticky bits that are inside a modern engine start to get brittle and start leaking fluids.
      • typically 8K miles for oil

        Must be an American car. Most European manufactures recommend oil changes at 16K or higher.

        • Really? Most quick searches I've seen show 8-10K miles; are you sure the 16K isn't for kilometers? That would be 10K miles - which is what VW recommends (at least in the US).
    • Most modern vehicles don't really have buttons or dials anymore, at least not in the traditional sense. Most of the buttons are "soft" not mechanical, even if they "click." Your dials and guages aren't connected directly to anything, they're just an analog stick moved by an electric signal.
    • Re: (Score:2, Informative)

      Come on, you take your eyes off the road all the time. Even with buttons and knobs.

      Anyway while driving you never touch the touchscreen in Tesla. All the crucial functions, steering, brake, turn signals, cruise on/off, wipers, head lamps, cruise control speed adjustment, media volume adjustment all have their own dedicated scroll wheels, switches on stalks etc.

      Voice command does answering phone, dialing out, change temperature or switch media sources.

      The huge screen shows the map, thats the only thing

    • i also prefer buttons/switches and dials plus i'd have to use my left hand to operate the screen and thats a task and half on an ordinary land based touchscreen
    • I come from a place that goes down to -40C in the winter and +35C in the summer. It breaks the seal on aluminum rims and it breaks welds over time. I pity anyone who attempts to own a Tesla in our climate.
    • by Osgeld ( 1900440 )

      bonus feature, they have no clue about long term testing and validation of electronics. when we did some stuff for them we asked them for their test specification and looked at us like retarded dogs ... Ford and GM have NYC sized phonebooks

  • by steveha ( 103154 ) on Wednesday June 24, 2020 @06:47PM (#60224416) Homepage

    Elon Musk thinks like a software developer, and Tesla did a lot of things differently than traditional car makers. This appears to be one case where it bit them.

    I think overall the "move fast and break things" approach has in fact worked out well for Tesla, especially since they are now slowing down and taking time more. The Model Y wasn't rushed at all; it's an evolved Model 3.

    • I always wondered what would happen if the Silicon Valley attitude to reliability and product development was applied to an industry with real warranties and real regulation. Thanks to Elon Musk, I now know the answer.
  • map data needs to be on an disk with big write limits. Not an smaller flash the is not build for daily changes

  • "prevents operators from being able to use the backup camera" Seriously? THIS is reason #1 why I don't like all of this technology inside an automobile. SOFTWARE can fail, and, when you have it making you dependent on your every operation of a vehicle, well, you panic. People are TERRIBLE drivers these days. Too many distractions.
    • by Anonymous Coward

      And yet we are supposed to embrace fully automated vehicles.

    • Convergence is a double-edged sword, and I for one think it should be avoided as much as possible, but no doubt it saves quite a bit of money to not have another whole module to handle the camera functionality somewhere else, and it certainly saves money to not have another screen.

      The real problem here is a lack of redundancy, which is sadly typical these days. Most vehicles (not just EVs) are built out of big expensive modules which are not practical to user-service. And they're still getting ridiculously

    • Having never had a backup camera until about a year ago... I'd argue they are a significant improvement over "using your mirror" and "looking over your shoulder". The field of vision is drastically improved, especially near the car.

      I still look over my shoulder while backing up, but it's a supplement to use of that camera.

      • Backup cameras are a necessity on a lot of newer vehicles because the visibility is so terrible. Thick pillars, gunslit windows, and high beltlines found on modern cars means without the camera you're practically blind backing up. A camera still provides better visibility than an older car that's designed so that the driver can see, but at least in that case if the camera failed you could still backup relatively safely.

        In my opinion, the stupidest part of most backup cameras systems is the screen is in th

        • Backup cameras are a necessity on a lot of newer vehicles because the visibility is so terrible. Thick pillars, gunslit windows, and high beltlines found on modern cars means without the camera you're practically blind backing up.

          An unfortunate side effect of better safety systems and more fuel efficient designs. With my 2015 Camry, the front and side visibility is still quite good, but the rear window is pretty useless for anything except seeing car-height or larger objects - there's no way you'd see anyone much under five feet tall who happened to be back there.

          Now my 93 Escort Wagon had a nice rear window. Also, if you backed into someone the car would probably fall apart, keeping the human safe.

          Per your other comment.. I would L

      • by jwdb ( 526327 )

        While a backup camera is better than nothing, I much prefer a beeping backup radar/sonar. That way I can both look and listen, and can be looking elsewhere than where the radar is sensing if needed. Only advantage the camera has over radar is that you can align to surface markings.

        It's a shame that they see to have gone out of style.

    • You know that you can still use that big piece of glass in the back of the car and mirrors to reverse, just like every other car for the last 50 years, right?

      If the backup camera fails, that's a service call to have someone fix it, like on any other car equipped with a reversing camera in the last 15 years. It should not cause the vehicle to be undrivable unless you are absolutely incompetent at operating a motor vehicle.

  • by 140Mandak262Jamuna ( 970587 ) on Wednesday June 24, 2020 @07:45PM (#60224608) Journal
    Apparently there is a debug log being written by Tesla into some flash memory. The log is excessive and it dumps so much of data that the flash memory read/write cycle limit gets exhausted and the chip starts failing.

    Apparently it is not an easy fix to replace the flash memory. If they could show this is systemic and happens to all the cars, then it would be a sort of defect. But if Tesla could show the amount of data being written will not exhaust the life of the flash memory within the warranty period, it would get off the hook.

    All the components have limited life in all the cars. Catalytic converter, oxygen sensor, struts and springs, even ball bearing and starter motors. Almost all these cars are out of warranty. And flash memory has lasted through warranty period, so it might not have legal liability to replace them.

    On the other hand, this is a premium automobile, Tesla is known for excellent customer service. More importantly its these early adopters who paid premium to fund the company so that I can have my Model 3. I am very sympathetic to the owners of early Tesla and I wish Tesla would do the right thing, even if they dont have to legally.

    • by drinkypoo ( 153816 ) <drink@hyperlogos.org> on Wednesday June 24, 2020 @09:13PM (#60224802) Homepage Journal

      if Tesla could show the amount of data being written will not exhaust the life of the flash memory within the warranty period, it would get off the hook.

      The camera is mentioned in the summary because it is a safety device. If the camera doesn't work then Tesla might reasonably be strong-armed into a recall even outside of the warranty period.

    • The problem is three fold:
      - the constant updates for the cars that are sent out, and they take up a good chunk of the 8gb eMMC -along with it's backup onboard. So what space there is available is constantly getting used. Even with wear leveling and so on..the units are hitting MTBF wayyyyy sooner!
      -with the new updates comes more functionality (videogames, karaoke, more automated driving software updates, and maps) and thus more read+writes to the eMMC. So it's a vicious circle. The third party repair folks

    • by AmiMoJo ( 196126 )

      Tesla is known for excellent customer service

      Tesla are known for terrible service. Early cars had a huge number of faults, later ones are getting better but they are still poorly made. Panel gaps of course but paint as well. Some people living in places with worse weather than California have been finding that the paint wears off within months around the wheel arches and door frames.

      None of this would be so bad if Tesla was good about fixing issues, but they are slow and always deny the problem and blame the customer first. Yes other manufacturers do

      • I have not owned earlier Teslas, so I can't say for sure.

        Either the service is least bad among its peers, or rest of the car is so good it makes up for poor service. The early Tesla owners were paying upwards of 80K or 90K for their cars. They are used to top of the line luxury cars. So they do have the experience to compare it with others luxury car makers. They dont own the car long enough to see paint peeling etc. They would not know stuff like panel gaps. But squeaks and rattles, service issues etc wou

        • by AmiMoJo ( 196126 )

          Tesla had their own Apple-style reality distortion field. Just listen to owners gush about their cars. Because of that some of them seem willing to put up with a lot of problems and just brush them off.

          Don't get me wrong, I adopted EVs fairly early and was ready to deal with issues. Fortunately I didn't have any except for a dodgy windscreen washer pump, but I was in the early adopter mindset. But there was also a limit and the issues many early Tesla owners had, often due to to poor service as much as defe

          • You are talking with personal experience. I am speculating and exrapolating based on tenuous media reports. So you are likely to be more correct than me in this issue.
    • But if Tesla could show the amount of data being written will not exhaust the life of the flash memory within the warranty period, it would get off the hook.

      Nope. Warranties don't work like that. Warranties cover manufacturing defects, not design defects. Design defects that dramatically alter the usable life of something from its expected life are usually covered under different "fit for service" laws.

      Not sure how this works in the USA, but the warranty period isn't relevant for this defect in many countries, the expected useful life is.

  • the same touch screen used in the Crew Dragon.
  • Here in Norway we have a Supreme Court decision that non-moving parts in consumer goods have to last at least 5 years, so I really hope my chip wears out before next spring when that 5-year period ends. Otherwise my best option is probably to dismount the board, send it off to Holland and have the chip replaced with something far larger (about $350), before it actually fails, since at that point my only option would be to pay Tesla for a full board replacement (about $2000).

    Terje

  • The endurance of an eMMC is in its datasheet. The write-load is something you always carefully control and measure (that is if you are competent). Seems to me somebody tried to have this designed on the cheap and now high cost of doing that becomes due.

  • So I live in a -40C to +35C climate. Interesting to know while the drive train is now bulletproof for any kind of weather, they made the dashboard equally fragile. The temperature swing will give a lot of people problems with solder joints after a few years.
  • Telsa appear to be having Quality Assurance issues here which they will discover will cost twice as much to fix as if they did things correctly the first time.
    Note:
    Quality Assurance you test at each stage of the manufacturing/assembly to make sure the component or part meets requirements and if mistakes are made you have a chance to fix them.
    Quality Control you test only at the end when the product has been made and if does not meet requirements you have to spend alot of money to fix the defect or throw t
  • The rear view camera not working is called a safety issue, despite that the three mirrors and rear windscreen are all perfectly usable. This is not a safety issue, this is a "would be nice to have, but not actually necessary" issue. They are required in 2018 and newer cars, but this is about 2013-2015 Teslas, they are not required equipment.

    There have been many class action lawsuits about failing infotainment systems which include the back up camera, involving the same model year or newer cars: Cadillac (20

  • Turn your head and look behind the car when backing up. So simple anyone can do it.

Keep up the good work! But please don't ask me to help.

Working...